I am having trouble with my pc in that it reboots itself.

I suspect a heat problem although I have a fan monitor and the case and CPU temps are fine. I have used a memory stress test program and it hasn't run into any problems. It doesn't seem to have any trouble just being left on but about 8 laps of Aston Nat is about the most it can manage without a crash.

I guess that leaves Graphic card over heating or power supply failing as the next most likely candiates?

Does anyone know how best to test either of these? Are there graphic stress test routines I can run to give it a good workout to see if that can reproduce the problem?

Any ideas would be appreciated.

Maxim

if the cpu and/or ram is a problem you should get errors trying to run Prime 95 'torture test' in blend mode.. if that runs overnight without errors the signs are good..

http://mersenne.org/gimps/p95v2414.exe

Do you have more games than LFS? Probably?.. See if you can play other games for long periods of time or do they also crash / hang / reboot?

If you have a version of 3dmark (even 2001 will do) see if you can loop that for a few hours..

Temperature is not often the problem. Only REALLY dust clogged systems or failed fans might cause issues; although of course one or two poorly assembled systems with loose coolers etc might make temperature a problem.

A detailed list of your pc components might help too, to see what we're talking about!

Try taking the case off your PC and running it. If it still does the same then I would say its not a heat isssue.

Also check your event log as that should have a load of warnings and errors at the time of the reboot that will tell you what happened.

I have to put my laptop up on bricks (Well game boxes really) to keek good airflow under it when I play LFS

what PSU have you got, the system specs and the graphics card

Quote from Woz :

Try taking the case off your PC and running it. If it still does the same then I would say its not a heat isssue

not neccsesarily, doing that can muck up the airflow and in some PC's actually make it hotter

Thanks for the feedback/suggestions.

I have downloaded 3dmark03 and prime test. So far it has run 3d mark 2 times without a problem with a score of around 14,800 if that makes a difference.
CPU temp is 35C, case is 34C and the hard drive with a temperature probe is at 40C. These are fairly typical temperatures and similar to when it reboots itself. I don't have a probe on the graphics card but I have checked the fan is turning. No dust problem yet as it is only about 6 months old since I assembled it.

System specs are
Athlon 64 - 3700
2 x 1 Gb DDR 400 Ram
ATI 1800XTX
Audigy 2 sound card
No other expansion cards
There are 3 hard drives and 2 dvds
It has 4 case fans attached to a fan controller.
The power supply is rated at 550W but is just a generic one.
It is running WinXP SP2

Other gamers installed ! Who has time? I do have Oblivion installed which does give my pc a good workout so I will let the Mrs know I am under instructions to play it for a few hours

How do I check the event log?

It has run faultlessly until a few weeks ago when this problem started happening. It runs fine when not stressed but 8 laps is about as far as I can get in LFS. Is it possible so update to windows has installed itself and is causing the trouble or would that be more likely to cause a BSD rather than a reboot?

I will let you know if it survives running firstly 3d mark and then prime over night.

Thanks again

Maxim

Ati tray tools will allow you to increase /decrease fan speed on your gfx card I use this tool and its great as the stock fan settings are not fab IIRC download here >> http://downloads.guru3d.com/download.php?det=733

SD.

It managed to run 3d mark 5 times before crashing. It rebooted itself and I ran it again but it managed 5 more without a problem.

It managed to run prime in blended stress test mode for 5 hours with no trouble.

CPU and case temps only rose about 1 or 2C

Using ATI tool suggested by SparkyDave I found that my graphics card has a temperature probe on it and it runs at about 60C. I overclocked it and it then ran at 72C even with a faster fan. I tested it with 3d mark again which was the second set of 5 I mentioned above. It had no trouble even at the higher temp so I thought "try LFS".

I managed about 22 laps this time before it rebooted. I am now suspecting the power supply. Does this sound likely?

Maxim

Quote from MaximUK :

It managed to run 3d mark 5 times before crashing. It rebooted itself and I ran it again but it managed 5 more without a problem.

It managed to run prime in blended stress test mode for 5 hours with no trouble.

CPU and case temps only rose about 1 or 2C

Using ATI tool suggested by SparkyDave I found that my graphics card has a temperature probe on it and it runs at about 60C. I overclocked it and it then ran at 72C even with a faster fan. I tested it with 3d mark again which was the second set of 5 I mentioned above. It had no trouble even at the higher temp so I thought "try LFS".

I managed about 22 laps this time before it rebooted. I am now suspecting the power supply. Does this sound likely?

Maxim

Welkl it could be PSU, the only thing that has me doubting it is tat you managed to overclock your gfx card and it ran fine.

Does the pc reboot when crashing or bluescreen? Disable the automatic restart on crash under Advanced system properties and when it crashed it should come up with some text and blue screen. Try to take a pic or write some stuff about the files and stop codes that caused the crash.

Event viewer can be found in Administator Tools in Control Panel and will let you know of any problems with the System, apps or security.

Graphics cards at 72c are no problem; most only 'throttle down' (I would call that 'braking' or 'slowing down ) at 120c. My card (6800gt) runs in the 90s and works fine.

Perhaps it is your PSU. Generic 550W models by cheap brands like QTEC are absolutely horrible. Surprisingly though they mostly work fine but they can cause issues. You're much better off with a 350W model of a good brand (Fortron, Seasonic..)

Perhaps you can borrow someones powersupply and see what happens with that? Or you can perhaps buy a new PSU (if decent brand, 350 or 400W is more than enough) and ask if you can get a refund if it doesn't solve the problem..

Edit: and it won't hurt trying to run Prime for more than 5 hours just to make sure cpu and ram are ok..

erm, usually the threshold is a little lower than 120c on Graphics cards and CPU's!

90 is too hot as well!

Whatever it is is a bit intermittent as I managed 3x10 lap races today with no reboot.

When it goes, it goes right back to reboot, no blue screen. I have changed the setting as suggested to not quto reboot.

I have ordered myself a coolermaster 620W. It wasn't much more than the 400W and I thought it would cover me from when I go Crossfireillepall

Maxim

Lets hope that solves it for you, I have a good feeling it will..

(Nvidia cards have their temperature threshold set at 120c which would indicate that they don't see a threatening situation until that point. There are 'passively cooled' cards that often run near 100c in most cases. As long as you can loop 3dmark overnight and you have no artifacts in games, it will last for many gaming years )

Quote from Niels Heusinkveld :

Perhaps it is your PSU. Generic 550W models by cheap brands like QTEC are absolutely horrible.

PARDON !!!! illepall illepall

I got Q Tec 550w
Quiet
Cool Running
Cheap
Powers My x1900xt and all the case lights/fans NO PROBLEM!

They don't supply enough voltage though, i had a 550W one and just wouldn't do it!

Really depends on how many amps it has on the 5v and 12v rails more than anything. The Thermaltake's lie about their wattage and are just rebranded generics so some generics can be ok as well. Totally depends on the rails, OFcourse the power might not be as clean as a good PSU though but it's good enough for a pc.

Software voltage monitoring is gimmicks. Get your multimeter out measure the voltage from one of the power molex connectors.

S14 drift, I continued with this:

"Surprisingly though they mostly work fine but they can cause issues. You're much better off with a 350W model of a good brand (Fortron, Seasonic..)"

So yeah they mostly work but are dodgy quality. Many times when reviewed and put under an actual load of 550W they die, sometimes taking pc parts with it. Plus often cheap PSUs, if they DO deliver the rated wattage, do that at room temperature.. Of course the components heat up under use and often the output drops to <50% at 50 degrees. Companies like Seasonic and Fortron tend to specify the load a 50 degrees or there abouts.. Anyways, I hope the Coolermaster (one step above 'cheap crap' mostly) will do the job MaximUK! Report back with the, hopefully, good news

I am going to get a Hiper Psu when i have money but this q-tec is good, keeps case temp low , has been reliable ..

Quote from Niels Heusinkveld :

...most only 'throttle down' (I would call that 'braking' or 'slowing down ) at 120c. My card (6800gt) runs in the 90s and works fine.

Quote from Niels Heusinkveld :

(Nvidia cards have their temperature threshold set at 120c which would indicate that they don't see a threatening situation until that point.

I absolutely cannot agree with that. Granted, my experience comes from a 6600GT, but while the 72°C you stated are no problem, 90°C are already on the edge. My card apparently had a crap stock cooler, so the temp went to 93°C max, with artifacts occuring at 90°C+. It didn't take long and the card broke down completely with a memory failure (random coloured pixels all over the place, instead of "just" malformed polygons caused by a failing vertex shader).

The 120°C nVidia sets as threshold is complete BS. You will never reach that temperature in normal conditions without your card breaking down first. At first I thought the same as you, hence I didn't see too much problems with the 93°C my card reached, but I soon learned otherwise - you'll also read the same when you look at some gfx card oriented forums. The catch is, the core itself might withstand temperatures of 120°C, but the surrounding parts can't. The only situation where you could get to that high temps is if you had perfectly functional cooling for all parts but the core itself, which is impossible (in the designs that I know of).